| The development of intelligent manufacturing has made machine vision more and more widely used in industrial robots,and has also put forward corresponding requirements for the application of machine vision in complex environments.More robots with monocular vision are used in current industrial applications,which can only obtain two-dimensional information of the target workpiece,and cannot cope with the complicated situation of workpiece posture diversity and stacking in actual working conditions.The research on manipulator positioning control system based on binocular vision was carried out in this paper.The main research contents of the thesis were as follows:(1)The acquisition of the position and orientation information of the stacked workpieces was studied.The internal parameters,external parameters and relative pose of two cameras were obtained through the double target experiment.The calibration results were used to geometrically transform and correct the image,The positioning effect of the camera on the stacked workpieces was studied based on the perspective deformable template matching and the rectangular pose evaluation method.On the basis of this,the three-dimensional pose of the workpiece was obtained by the perspective deformable template matching method,and the depth information obtained by the stereo matching method was used to correct the foregoing pose,as the exact pose of the target workpiece in the camera coordinate system.(2)In order to verify the effectiveness of the research method and algorithm,a cantilever four-axis manipulator experimental platform was constructed.The system architecture of “PC+motion control card” was adopted for the experimental platform control system,and the system was designed and constructed by using IMB-750 GC industrial camera of Korea IMI Company and MT-263 motion control card of Shanghai Xiaomo Company.(3)The control system programs were developed to facilitate experimentation and application.The control system man-machine interface includes functions such as communication,robot motion control,image processing,and workpiece positioning grab and placement.(4)Using the developed experimental platform,the research and experimental analysis on the system acquisition of position and posture information of stacked workpieces and automatic grasping workpiece by manipulator were carried out.The relationship was deduced to transform the position and posture obtained from image positioning into the three-dimensional position and posture of the target workpiece in the basic coordinate system of the manipulator,which was studied in the process of grabbing.The inverse kinematics solution was combined with the manipulator structure to obtain the grabbing trajectory data of the manipulator.Aiming at the classification of locatable and non-locatable targets in stacked objects,the template matching process was optimized by setting the stacking coefficient value and the matching score threshold to implement classified grabbing.The grabbing order of“from top to bottom,from left to right,from back to front” was formulated as the evaluation criterion for the priority of grabbing,and the grabbing efficiency was improved by researching the optimization of grabbing trajectory planning;In view of the illumination change and the interference of various workpieces,a series of experiments were carried out to compare and analyze the success rate of grabbing and positioning accuracy of stacked workpieces,which verifies the feasibility of the system and the adaptability to the complex environment,and the reasons for the grabbing errors were analyzed.The results of grabbing experiments show that the system developed in this paper has a good positioning and recognition effect for stacked workpieces in the complex environment,and the grabbing accuracy can meet the requirements of common industrial manipulators for grabbing workpieces.Relevant research results can be used for reference to promote the application of industrial robots in intelligent production systems. |
PDF Full Text Request